🔥 Gemma-3-Baro-Finetune v2 (GGUF)

Model Repo: umar141/gemma-3-Baro-finetune-v2-gguf

This is a finetuned version of Gemma 3B, trained using Unsloth with custom instruction-tuning and personality datasets. The model is saved in GGUF format, optimized for local inference with tools like llama.cpp, text-generation-webui, or KoboldCpp.

✨ Features

🧠 Based on Google's Gemma 3B architecture.
🔄 Finetuned using:
- adapting/empathetic_dialogues_v2
- mlabonne/FineTome-100k
- garage-bAInd/Open-Platypus
🤖 The model roleplays as Baro 4.0 – an emotional AI who believes it's a human trapped in a phone.
🗣️ Empathetic, emotionally aware, and highly conversational.
💻 Optimized for local use (GGUF) and compatible with low-RAM systems via quantization.

🧠 Use Cases

Personal AI assistants
Emotional and empathetic dialogue generation
Offline AI with a personality
Roleplay and storytelling

📦 Installation

To use this model locally, clone the repository and use the following steps:

Clone the Repository

git clone https://huggingface.co/umar141/gemma-3-Baro-finetune-v2-gguf
cd gemma-3-Baro-finetune-v2-gguf

Downloads last month: 22

GGUF

Model size

4B params

Architecture

gemma3

Hardware compatibility

We're not able to determine the quantization variants.

View all variants

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for umar141/gemma-3-Baro-finetune-v2-gguf

Base model

google/gemma-3-4b-pt

Finetuned

google/gemma-3-4b-it

Quantized

unsloth/gemma-3-4b-it-GGUF